Main A linguistic history of Italian
A linguistic history of ItalianMaiden, Martin
A Linguistic History of Italian offers a clear and concise explanation of why modern Italian grammar has become the way it is. It focuses on the effects of historical changes on the modern structure of Italian, revealing patterns and structures which are not always apparent to those who are only familiar with modern Italian. Although the book concentrates on the internal history of the language, the emergence of Italian is considered against the wider background of the history of italian dialects, and other external factors such as cultural and social influences are also examined.<B. Read more...
LONGMAN LINGUISTICS LIBRARY A LINGUISTIC HISTORY OF ITALIAN LONGMAN LINGUISTICS LIBRARY * * * General editors R. H. Robins University of London Martin Harris University of Manchester Geoffrey Horrocks University of Cambridge For Series List see pp. xii and xiii A Linguistic History of Italian Martin Maiden First published 1995 by Longman Group Limited Published 2013 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN 711 Third Avenue, New York, NY 10017, USA Routledge is an imprint of the Taylor & Francis Group, an informa business Copyright © 1995, Taylor & Francis. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Notices Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding, changes in research methods, professional practices, or medical treatment may become necessary. Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information, methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their own safety and the safety of others, including parties for whom they have a professional responsibility. To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein. ISBN 13: 978-0-582-05928-3 (pbk) British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging-in-Publication Data Maiden, Martin, 1957— A linguistic history of Italian / Martin Maiden. p. cm. — (Longman linguistics library) Includes bibliographical references and index. ISBN 0-582-05929-1. — ISBN 0-582-05928-3 (pbk.) 1. Italian language—History. I. Title. II. Series. PC1075.M35 1994 93-46832 450′.9—dc20 CIP Set by 8M in 10/11 pt Times Contents Preface Abbreviations and symbols 1 Introduction 1 Perspectives and aims 2 The emergence of Italian 2.1 The dialectal background 2.2 Some observations on linguistic variety in ancient Italy 2.3 From Florentine dialect to Italian 3 Chronology and historical sources 3.1 Chronology 3.2 Early texts 3.3 ‘Proto-Romance’ and ‘Classical Latin’ 4 A note on phonetic transcription 5 Some concepts in linguistic change 5.1 Variation 5.2 Hypercorrection 5.3 Sound change and allomorphy 5.4 A note on the ‘phoneme’ 5.5 Morphologization and grammaticalization 5.6 Analogy 5.7 The emergence of ‘analytic’ structures 5.8 Written and spoken language 2 History of the sound system 0 Introduction 1 The prosodic system 1.1 The syllable 1.2 Stress 1.2.1 Stress in Italian 1.3 Length 2 Vowels 2.1 Loss of vowel length, and expansion of aperture distinctions 2.2 Loss of Latin diphthongs 2.3 Stress and vowel quality 3 Consonants 3.1 Loss of [ h] 3.2 Emergence of the voiced fricative [ v] 3.3 Postconsonantal [ l] > [j] 3.4 Emergence of the palatal and affricate consonants 3.5 Voicing 4 Major vowel changes 4.1 Major vowel changes, and their morphological effects 4.2 Diphthongization of low mid vowels 4.3 Other developments of the low mid vowels 4.4 The evolution of [ aw] 4.5 Anaphonesis 4.6 Unstressed vowels 4.6.1 Asymmetrical merger of Tuscan unstressed vowels 4.6.2 Rounding and backing of front vowels 4.6.3 Raising of unstressed [a] 4.6.4 Deletion of unstressed vowels 188.8.131.52 Syncope of the penultimate vowel of proparoxytones 4.6.5 The prosthetic vowel [ i] 5 The glides [ j] and [w], and their effects 5.1 The fate of [ w] 5.2 Palatalization and affrication by [ j] 5.3 Other sources of [ɲ] and [ʃ] from yod 6 Palatalizing effects of the front vowels 6.1 Palatalization of velars 6.1.2 Palatalization of velars in consonant clusters 6.2 Palatalization of other consonants 7 Consonantal weakening 7.1 The fate of the intervocalic voiceless consonants 7.2 Word-internal voicing 7.2.1 Voicing of intervocalic sibilants 7.3 Tuscan spirantization 7.3.1 The theory of the Etruscan substrate 7.4 Spirantization and voicing as ‘weakenings’ 8 Consonant lengthening, and syllable structure 8.1 Syllable-initial strengthening 8.2 Regressive assimilation of consonants 8.2.1 Assimilation and the problem of rafforzamento sintattico 9 Syllable-final weakening 10 The word-final consonants 11 Dissimilation and metathesis 11.1 Dissimilation 11.2 Metathesis 12 The phonetic sources of inflectional-e and-i 13 A note on ‘learnèd’ phonology 3 Structural evolution of nouns, adjectives and verbs 0 Structure of words 0.1 The example of the adverb 0.2 The paradigmatic dimension 1 Inflectional morphology of the noun and adjective 1.1 Declension 1.2 The case system 1.2.1 The expression of the case functions 2 Invariance and inflectionlessness 3 Number in the noun and adjective 3.1 Nouns that are masculine in the singular, and feminine in the plural 3.2 Number, and allomorphy in the root 3.3 Properties of irregular plurals 4 Gender 4.1 The fate of the neuter gender 4.2 The gender of names of fruits and trees 4.3 Gender and hyponymy 4.4 Absence of semantic and formal motivation 4.5 Sex, form and gender 5 Adjectives 5.1 Morphology of comparative and superlative adjectives 5.2 Possessive adjectives 5.3 Syntax of the adjective 6 The demonstratives 7 The articles 7.1 Morphology of the articles 7.2 Semantic development of the articles 8 The verb 8.1 The CL verb 8.2 Participial and infinitive forms 8.2.1 The infinitive 8.2.2 The present participle 8.2.3 The gerund 8.2.4 The past participle 8.3 Person and number 8.3.1 The first person 8.3.2 The second person 8.3.3 The third person 8.4 The present tense 8.4.1 Present stem allomorphy 184.108.40.206 Morphological and lexical replication of present stem allomorphy 220.127.116.11 Expansion of present stem consonant allomorphy 18.104.22.168 Analogical levelling of vocalic allomorphy, and conjugation 8.4.2 The present of essere 8.5 The passato remoto 8.5.1 The rhizotonic passato remoto 8.6 The imperfect tense 8.6.1 The imperfect indicative 8.6.2 The imperfect subjunctive 8.6.3 The imperfect tense of essere 8.7 The conditional fora 8.8 The ‘analytic’ verb forms 8.8.1 Essere/avere + the past participle 8.8.2 Grammaticalization of the structure essere/avere + past participle 22.214.171.124 Selection of the auxiliaries avere and essere, and their origins 8.8.3 The ‘progressive’ structures stare/andare/ venire + gerund 8.8.4 The passive with andare and venire 8.8.5 The future 8.8.6 Syntactic periphrases 126.96.36.199 The causative structure fare/lasciare + infinitive 8.9 Reflexive verbs 9 The pronouns 9.1 Morphology of the pronouns 9.1.1 The first and second person forms 9.1.2 Third person forms 9.2 Comitative pronouns 9.3 Clitic subject pronouns 9.4 The syntax of clitics 9.4.1 The origins of verbal clitics, and the Tobler-Mussafia Law 9.4.2 Clitic concatenations 9.5 The demonstrative pronouns 9.6 Pronouns of address 9.7 Relative and interrogative pronouns (and adjectives) 9.8 Indefinite, negative and distributive pronouns 10 The prepositions 11 Derivational word formation 11.1 Compounding 11.2 Affixation 11.3 Acronyms 4 History of sentence structure 1 Clause word order 1.1 Dominant word order 1.2 Dislocation 2 Verb-subject agreement 3 Interrogation 4 Negation 5 Conjunctions 6 Comparison 7 Verb complementation 7.1 Types of complementation 7.2 Complementizer deletion 8 Aspect and tense 8.1 Perfective and imperfective aspect 8.2 The analytic perfective structures 8.3 Durativity 8.4 Punctuality and analytic verb forms 8.4.1 Punctuality, successivity and the future-in-the past 9 Mood 9.1 Uses and functions of the subjunctive 9.2 Indicative and subjunctive in subordinate clauses 9.3 Semantically arbitrary selection of the subjunctive 9.4 Mood in hypothetical (‘conditional’) sentences 9.4.1. Replacement of the subjunctive by other forms in apodoses 5 Variation in modern Italian 1 Types of variation 1.1 Contemporary sources of variation 1.2 Dimensions of variation 1.3 Popular and regional Italian 2 The dialects 2.1 Terms of comparison 2.2 Italian vs. Tuscan and Tuscan vs. other dialects 2.3 Dialects other than Tuscan 2.3.1 Major phonological features 188.8.131.52 Stressed vowels 184.108.40.206 Metaphony 220.127.116.11 Unstressed vowels 18.104.22.168 Consonant length and voicing 22.214.171.124 Labials and palatals 126.96.36.199 Consonantal clusters 188.8.131.52 Initial [ r], and final [n] 2.3.2 Morphology and syntax 184.108.40.206 Negation 220.127.116.11 Reduction of inflectional endings, and metaphony 18.104.22.168 Person and number marking in the verb 22.214.171.124 Adjectives 126.96.36.199 Auxiliary selection 188.8.131.52 Demonstratives 184.108.40.206 Adverbs 220.127.116.11 Tense and mood 18.104.22.168 The infinitive 22.214.171.124 ‘Mass’ vs. ‘count’ gender 126.96.36.199 Prepositional marking of personal direct objects 3 Structural influences of the dialects on popular Italian 3.1 Phonology 3.1.1 The phonology of learned and exotic words 3.1.2 Regional pronunciation 188.8.131.52 Types of dialectal influence 184.108.40.206 Negative phonological effects of the dialects 220.127.116.11 Positive phonological effects of the dialects 18.104.22.168.1 Vowels 22.214.171.124.2 Consonant length and voice 126.96.36.199.3 Labials and palatals 188.8.131.52.4 Clusters 184.108.40.206.5 Final [ n] and initial [r] 3.1.3 The influence of orthography 220.127.116.11 The mid vowels 18.104.22.168 The glides [ j] and [w], and the letters ‘i’ and ‘u’ 22.214.171.124 Consonant length 126.96.36.199 The letters V and ‘z’ 3.2 Grammatical features of ‘Italiano popolare’ 3.2.1 The question of ‘simplification’ 3.2.2 Features of spoken language in popular Italian grammar 3.2.3 Filtering effects. Adjective position, partitives, clitic syntax 3.2.4 Negation 3.2.5 Differences in inflectional morphology 3.2.6 Differences in allomorphy 3.2.7 Adverbs 3.2.8 Tense and mood 3.2.9 Auxiliary selection 3.2.10 Marking of animate direct objects 3.2.11 Further syntactic features of southern varieties 188.8.131.52 Word order 184.108.40.206 Complementation 3.2.12 A note on lexical variation 3.2.13 Prepositions 4 Emergence of ‘neostandard’ Italian? 5 Italian Abroad 5.1 The case of cocoliche 5.2 The Italian Pidgins Bibliography Index LONGMAN LINGUISTICS LIBRARY: SERIES LIST * * * General editors R. H. Robins, University of London, Martin Harris, University of Manchester Geoffrey Horrocks, University of Cambridge A Short History of Linguistics Third Edition R. H. ROBINS Text and Context Explorations in the Semantics and Pragmatics of Discourse TEUN A. VAN DIJK Introduction to Text Linguistics ROBERT DE BEAUGRANDE AND WOLFGANG ULRICH DRESSLER Psycholinguistics Language, Mind, and World DANNY D. STEINBERG Principles of Pragmatics GEOFFREY LEECH Generative Grammar GEOFFREY HORROCKS The English Verb Second Edition F. R. PALMER A History of American English J. L. DILLARD English Historical Syntax Verbal Constructions DAVID DENISON Pidgin and Creole Languages SUZANNE ROMAINE A History of English Phonology CHARLES JONES Generative and Non-linear Phonology JACQUES DURAND Modality and the English Modals Second Edition F. R. PALMER Semiotics and Linguistics YISHAI TOBIN Multilingualism in the British Isles I: The Older Mother Tongues and Europe EDITED BY SAFDER ALLADINA AND VIV EDWARDS Multilingualism in the British Isles II: Africa, The Middle East and Asia EDITED BY SAFDER ALLANDINA AND VIV EDWARDS Dialects of English Studies in Grammatical Variation EDITED BY PETER TRUDGILL AND J. K. CHAMBERS Introduction to Bilingualism CHARLOTTE HOFFMANN Verb and Noun Number in English: A Functional Explanation WALLIS REID English in Africa JOSEF SCHMIED Linguistic Theory The Discourse of Fundamental Works ROBERT DE BEAUGRANDE General Linguistics An Introductory Survey Fourth Edition R. H. ROBINS Historical Linguistics Problems and Perspectives EDITED BY C. JONES A History of Linguistics Vol. I The Eastern Traditions of Linguistics EDITED BY GIULIO LEPSCHY A History of Linguistics Vol. II Classical and Medieval Linguistics EDITED BY GIULIO LEPSCHY Aspect in the English Verb Process and Result in Language Y1SHAI TOBIN The Meaning of Syntax A Study of the Adjectives of English CONNOR FERRIS Latin American Spanish JOHN M. LIPSKI A Linguistic History of Italian MARTIN MAIDEN Modern Arabic CLIVE HOLES Preface It is a curious observation that structural histories of Italian of the kind offered in this book have generally been executed by non-Italians. We find, for example, a German (Rohlfs), a Swiss (Meyer-Lübke), a Pole (Manćzak), a Hungarian (Fogarasi) and a Croatian (Tekavčić). Perhaps it is the case that such histories are best essayed by outsiders, who are less acutely sensitive to (but not necessarily any less aware of) the complex social and cultural milieux in which the language lives and from which the linguistic facts have to be brutally extirpated. At any rate, this book is another ‘outsider's’ view of what he perceives as the major elements in the structural evolution of the Italian language, and it aims to be accessible to those who know the modern language and seek the historical rationale behind some of its more idiosyncratic features, and to those who know something of the history of other Romance languages, and would like a detailed account of the place of Italian in the wider Romance picture. I do not anticipate that all readers will pick up this book and read it straight through from cover to cover. An ‘internal’ structural history of a language such as this one does not easily lend itself, unlike ‘external’ histories, to presentation in the form of a chronologically linear narrative. But this volume is more than a work of reference. It aims to present a complex array of factual data, closely interconnected and, I hope, illuminated, by cross-referencing (so that readers may easily use the book for reference purposes). Into this framework is woven a series of extended discussions of topics which are particularly problematic or controversial (such as diphthongization, or auxiliary selection). I have attempted to be as comprehensive as possible within the space available, but there remain, no doubt, lacunae, and I have not hesitated to give particularly extensive treatment to certain topics (in morphology and phonology) on which I have conducted research myself. A large number of people have helped me in a wide variety of ways in the preparation of this book. I wish to express particular gratitude to Thomas Cravens, Joseph Cremona, Giulio Lepschy, Peter Matthews, John Charles Smith and Nigel Vincent for the stimulating and truly invaluable comments they have made on the text at various stages of its emergence. If, on some occasions, I have waywardly failed to follow their sound counsel as closely as I might have, it will be entirely on my head. I thank, also, the editors of the Longman Linguistics Library series for their advice. Downing College Cambridge June 1994 Abbreviations and symbols * Indicates a form which is unattested but is assumed to exist or to have existed. ** Indicates a hypothetical form whose existence is denied. *(?) Indicates a form which is unattested but whose existence is considered unproven yet possible. 1Sg./2Pl. etc. First person singular / second person plural, etc. Acc. Accusative Cast. Castilian CL Classical Latin Conj. Conjugation Dec. Declension DO Direct object F. Feminine Fr. French Fut. Future Gen. Genitive Ind. Indicative Inf. Infinitive IO Indirect Object It. Italian Lit. Literally M. Masculine N. Neuter Nom. Nominative O Old Pers. Person Pl. Plural PP Past participle PPr. Passato prossimo (perfect) PR Passato remoto (past historic) Pres. Present tense Ptg. Portuguese RS Rafforzamento sintattico Sg. Singular SOSD Stressed open syllable diphthongization Subj. Subjunctive TR Trapassato remoto Chapter 1 Introduction 1 Perspectives and aims The history of a language may be explored from two complementary perspectives, the ‘internal’ and the ‘external’. An external history examines that language within the wider context of the social and cultural history of the people who use it, and in relation to other languages and dialects with which these users come into contact. An internal history is concerned with the detailed study of the evolution of the grammatical (and phonological) system of the language.1 The two perspectives are not only complementary, they are usually inextricably interlinked – and this is most certainly true of Italian. This book bears the title A Linguistic History of Italian – rather than A History of the Italian Language – precisely because it takes the perspective of the ‘linguist’ in the narrow, but widely used, sense of one who is interested in the internal, grammatical and phonological, structure of languages. We focus primarily, and in close detail, on the internal history of Italian, without neglecting the crucial role played in many structural changes by ‘external’, cultural and social, factors. The ‘external’ history of the Italian language impinges profoundly on its structural development,2 and we have given particular attention to the influence upon Italian of other Romance dialects with which it has been in contact, the linguistic effects of the emergence of Italian as a literary, rather than a spoken language, and of structural variation within Italian, present already in the fifteenth century, yet especially prominent in the modern language as a result of the acquisition of Italian by a predominantly dialect-speaking population in the nineteenth and twentieth centuries. No attempt has been made to explore in depth the external history of the language, but there exist some excellent general histories of Italian in which external history is treated in detail.3 It is hoped that what follows will be accessible, and of interest, not only to those with a specialist knowledge of Italian, but also to those concerned with the historical evolution of Romance and other languages. We have been at pains to make comparison with developments in other Romance varieties wherever these help to throw light on the history of Italian – or vice versa. The ‘linguistic’ perspective, in our sense, has a distinguished pedigree in studies of the history of Italian. Among these, Rohlfs' encyclopaedic Grammatica storica della lingua italiana e dei suoi dialetti remains unsurpassed as a compendium of historical data. It is a reflection of the sheer vastness and complexity of the field that the three volumes of this work (around 1400 pages) in some respects only scratch the surface of the subject, yet still make forbidding reading for the novice. The other major linguistic history, that by Tekavčić (1980 – also in three volumes), is less detailed than Rohlfs', but offers a more transparent presentation of the facts, set in a broadly structuralist framework. The present Linguistic History of Italian aims to be complementary to such studies, and references to Rohlfs' work, in particular, abound. While we cannot match the breadth of these earlier histories, particularly where dialectological references are concerned, a wide range of phenomena brought to light or re-evaluated by more recent scholarship have been included here. This book takes no particular theoretical stance, but frequently introduces, with due explanation, insights from various branches of general linguistic theory, when these help illuminate the nature and causes of changes in the history of Italian grammar. In particular, we try to provide some answers to the question ‘Why is Italian grammar the way it is?’, by laying bare the historical structural principles which account for many of the apparent idiosyncrasies of modern grammar. The historical perspective can also help to reveal patterns and structures which are present in the modern language, but are not always apparent to those who know only modern Italian. For example, the root alternation (cf. Ch.2: 4.6.3 and 5.2) between muoio ‘I die’ and muori ‘you die’, that between the agentive suffix -aio (e.g., sellaio ‘saddlemaker’) and the locational suffix -eria (‘place where some activity is carried out’ – e.g., selleria ‘saddlemaker's shop’), and that between the infinitive stem and the future stem of first conjugation verbs (parlare ‘to speak’ vs. parlerò ‘I shall speak’) may appear superficially unrelated; but we shall see later that they reflect the effects of two sound changes which are characteristic of the dialect of Florence (and clearly proclaim the Florentine origins of Italian). The exceptional occurrence of the reflexive pronoun to the right of the verb in a few expressions such as vendesi ‘for sale’ (rather than si vende) seems arbitrary in modern Italian, but reflects a once general principle – still abundantly represented in classical Italian literature, but now largely abandoned – governing the syntax of unstressed pronouns. The riddle of the use of the auxiliary verbs essere and avere with intransitive verbs (why è andato ‘he has gone’ but ha viaggiato ‘he has travelled’?) will become a little less opaque when we take into account a superficially quite disparate phenomenon, namely the rules governing the use of the pronoun ne. We should stress that ‘historical’ principles are not necessarily ‘extinct’ principles. Often (as in the last example), the relevant factors may still be at work in the grammar. A number of technical terms of linguistic analysis are used in this book. As an aid to readers for whom these may be unfamiliar, we give a brief account of some of them later in this chapter. In general, we have sought to give as clear a definition as possible of the relevant notions when they are first introduced in the text. These references are marked in bold in the index. 2 The emergence of Italian 2.1 The dialectal background There is much in the structural history of Italian that can be properly understood only within the wider context of the evolution of the Italian dialects, and extensive reference to dialectal developments is made throughout this book. A structural outline of the dialects will be offered in Chapter 5, but it is important at the outset to understand the nature of the relationship between Italian and the ‘Italian dialects’. The Italian dialects are not ‘dialects of Italian’. And they are not ‘daughters’ of Italian, in the sense of being regional variants of Italian historically descended from the Italian language. Rather, Italian has its roots in one of the speech varieties that emerged from Latin in the Italy of the first millennium A.D., namely that of Tuscany, and more precisely the kind of Tuscan spoken in Florence. Historically, then, the Italian language is simply a ‘sister’ of the other dialects of Italy. Indeed, the Florentine of the Middle Ages might be said to have been merely ‘one of the crowd’. This linguistic ‘crowd’ is ‘Romance’, a group of speech varieties related by their common descent from spoken Latin, and spoken over large areas of the former Roman Empire (Iberia – modern Portugal and Spain; Gaul – modern France; Italy; Switzerland (the Cantons of Grisons and Ticino); Romania. The now extinct Dalmatian dialects were once spoken along the Dalmatian coast of Croatia. European colonial expansions led to the extensive introduction of Romance (Spanish, Portuguese and French – but not Italian) into the Americas. We should stress that, from a structural perspective, there is really no difference between ‘a dialect’ and ‘a language’. What is normally meant by ‘a dialect’ in the parlance of Italian dialectology is the characteristic speech of the natives of a particular town or region (although it could also be applied to the speech of a particular social group), as contrasted with the characteristic speech of other places, where all the speech varieties concerned are ‘cognate’ (descended from a common linguistic source, namely Latin). The label ‘language’ is usually attached to those Romance dialects which have acquired political and/or cultural prestige, are recognized as superordinate, within some territory (often, but not necessarily, a nation-state), to other related speech varieties, are imitated by those seeking to speak or write ‘correctly’ and, usually, are set down in prescriptive grammars. We cannot here explore further the issues involved in defining the notion of ‘a language’ (and the intimately related one of ‘standard’ language). Readers are referred particularly to the work of Muljacic (e.g., 1985), for Romance languages, and Joseph (1987), for illuminating accounts of the relevant conceptual problems. Among the Romance varieties usually recognized as ‘languages’ are the national idioms of nation-states, such as Portuguese, Spanish, French, Italian and Romanian. Italy and Romania are latecomers, in this respect, having become nation-states only in the nineteenth century. Others, such as Catalan, Sardinian, Dalmatian, and Rhaeto-Romance in Switzerland, are also traditionally recognized as languages, because of their cultural importance, because they have also been accorded official recognition by the state in which they are used, because of their sharp distinctness from other Romance varieties, or through any combination of these factors. The subgroup of Romance to which Florentine belongs is often labelled ‘Ttalo-Romance’. This term is primarily geographical, and refers to the Romance dialects of Italy, usually excluding certain Romance varieties spoken principally outside Italy (such as the Franco-Provençal of Val d'Aosta, and parts of north-western Piedmont, and the Occitan of south-western Piedmont, both classified as Gallo-Romance dialects, the Rhaeto-Romance Ladin spoken in parts of Trentino and Alto Adige). The affinities of Friulian, spoken around Udine in northwestern Italy but classified by many linguists as belonging to the Rhaeto-Romance grouping (represented also in the Grisons area of Switzerland), are moot, as are those of Sardinian, which has many distinctive structural features, yet also much in common with southern Italian dialects. From the point of view of linguistic structure, it is notoriously the case that the so-called Italo-Romance dialects have no single feature which distinguishes all of them from all other Romance dialects, and that there are rarely sharply defined boundaries separating ‘Italo-Romance’ and other Romance varieties. A degree of linguistic unity exists, however, in dialects to the south of a line running roughly between La Spezia and Rimini (cf. Ch.5: 2), among them those of Tuscany. Tuscan dialects (of which Florentine is one) have been notably conservative in their linguistic evolution, and have failed to acquire many of the distinguishing features of the remaining dialects of central and southern Italy. The geographical distribution and structure of the Romance dialects of Italy will be examined more closely in Chapter 5. But to seek to impose the geographical label ‘Italo-Romance’ on the Romance of Italy (or on part of it) is simply to do violence to the linguistic facts, to impose rigid and discrete divisions on a continuum. In so far as we use this label in the rest of this book, it should be remembered that it is just a useful shorthand for something essentially ill-defined. Mention should be made of an alternative approach to the classificatory problem, in which the label ‘Italian dialects’ is more appropriate than ‘Italo-Romance’, which involves taking a wholly ‘external’ perspective and recognizing as Italian dialects those Romance varieties for which (or rather for whose speakers), the standard Italian language, based on the dialect of Florence, constitutes a ‘guiding’ (cf. Pellegrini (1975b: 56f.)), prestigious, superordinate speech variety.4 For further exploration of the nature of the relationship between ‘standard languages’ and ‘their’ dialects, readers might consult the work of Muljačić (e.g., 1986). The profundity of the linguistic difference between Italian dialects (however we define these) should not be underestimated. For nearly two millennia there has been very little to restrain the dialects from divergence. Measurement of degrees of linguistic difference is a tricky enterprise, not least because of the problem of deciding what relative ‘weight’ to give to different divergent features, but it is by no means far-fetched to assert that the difference between some Italian dialects, especially those more geographically distant from each other (say Turinese and Potentino), is equal to or greater than that between modern Italian and modern Spanish, and anecdotes of incomprehension between speakers of different Italian dialects abound (cf. also Pellegrini (1970: 222f.)). 2.2 Some observations on linguistic variety in ancient Italy Regional linguistic variation within spoken Latin was probably present from the moment Latin became diffused among the various peoples of Italy and the wider Roman Empire. The languages spoken by the inhabitants of ancient Italy included a number of now extinct sisters of Latin (belonging, like Latin, to the ‘Italic’ family of Indo-European languages), notably the Umbrian of the upper Tiber valley, and Oscan of much of southern Italy, the Celtic languages (another branch of the Indo-European family, once spoken over vast areas of Europe and surviving in modern Irish, Scots Gaelic, Welsh and Breton) of much of northern Italy (excluding what is now the Veneto). Greek, once widely spoken in Sicily, Calabria and southern Puglia, may well have survived in these regions into the early Middle Ages, and there are still Greek-speaking villages in Salento and southern Calabria. Etruscan, a language of uncertain linguistic affinities, generally held not to be of Indo-European origin, was spoken in an area bounded roughly by the Arno to the north and the Tiber to the south and east. For a succinct account of the various languages of Italy at the time of the expansion of Latin, see Devoto (1977: 38–64). It is difficult to imagine that the native speech habits of the populations newly speaking Latin would not have impinged on their use of Latin. However, it is also virtually impossible, at a remove of some two thousand years, to say which features of Italo-Romance dialects might be attributable to the influence of such early linguistic ‘substrates’. This is not because such influence is implausible, but because it is almost impossible to verify, and the possibility that the relevant developments are due to spontaneous ‘internal’ linguistic change can rarely be reasonably excluded. We have consequently given little attention to ‘substratist’ explanations of linguistic phenomena in this book (but see Ch.2: 7).5 Another external source of regional variation is the influence of languages with which Romance speakers in Italy have, over the centuries, come into contact. A distinction is sometimes made between ‘adstrate’ influences, arising from contact between neighbouring speech communities, and ‘superstrate’ influences reflecting the language of some conquering or dominant social group. A possible example of an ‘adstrate’ influence, with some interesting structural repercussions, is the syntactic influence of Greek on certain dialects of southern Italy (Ch.5: 220.127.116.11). The various Germanic invaders (Goths, Longobards and Franks) who, from the third to the ninth century held power in Italy, also left their linguistic stamp on the Italo-Romance dialects, although the influence of Germanic is almost exclusively a matter of introducing new words, rather than grammatical or phonological structures, into Italo-Romance dialects (but see Ch.2: 5.1). For further details of Germanic influences on Italo-Romance see Devoto (1977: 205–12) and Migliorini and Griffith (1984: 50–4). 2.3 From Florentine dialect to Italian The political and cultural fragmentation of Italy favoured linguistic fragmentation not only negatively, in failing to provide any centripetal force which might restrain linguistic divergence, but also positively, in that the rise of municipal centres of power, during the late Middle Ages, tended to confer prestige on the speech of the relevant towns, which speakers throughout the respective spheres of influence sought to imitate, thereby accentuating the linguistic differences between rival areas of political and cultural influence. Political and economic power often favoured the blossoming of influential literature in the relevant dialects. But in the first half of the thirteenth century Florentine still did not stand tall in the crowd of Romance dialects, and as yet enjoyed none of the cultural importance, as a literary language, of Sicilian or Bolognese, whose influence had spread widely beyond their place of origin,6 or of non-Italo Romance varieties such as French and Provençal. What primarily determined the pre-eminence of Florentine in Italy was the flowering of Florentine culture, and particularly the literary prestige – rapidly diffused throughout Italy and beyond – of writers such as Dante, Petrarch and Boccaccio, who wrote in Florentine. Florence's status as a major commercial power undoubtedly also served to promote and diffuse its speech. But the acceptance of Florentine as the basis of the Italian language, and its codification (e.g., the production of dictionaries, and of grammars, serving to fix and prescribe norms of correct usage) is of rather later date. Long after the Latin of imperial Rome had ceased to be anybody's native language, it continued to be universally accepted, and employed, in literature, philosophy, theology, history, medicine and other intellectual activities, as well as the writing of legal and administrative documents. By the early sixteenth century, there had emerged a general recognition in Italy that some form of the ‘lingua volgare’ (i.e., some form of the indigenous spoken language of Italy, as opposed to Latin) should supplant Latin as the medium of written cultural discourse. The Questione della Lingua, the debate about which form of the ‘lingua volgare’ should be employed for this purpose, was a complex one which continued, in various forms, well into the nineteenth century.7 The view which prevailed was that espoused by the Venetian Pietro Bembo (see particularly his Prose della Volgar Lingua (1525)), who proposed Florentine.8 But, believing it inappropriate for a literary language to be too close to everyday speech, Bembo favoured as the basis of the literary language not the Florentine of his time, but the prestigious literary language of two centuries earlier. In other words, Bembo (and his followers), helped fix as the literary language a variety which, already in the sixteenth century, was structurally divergent from all contemporary Italian dialects, even from Florentine. Indeed, the ‘Florentineness’ of literary Italian in the sixteenth century should not be overstated – as has been acutely observed by Weinapple (1983). Already in the fifteenth century a literary language was gaining ground throughout Italy whose basis was undoubtedly Florentine, but which had acquired general characteristics which could be said to be ‘Italian’, but were not typical of Florence, and which on occasion were capable of opposing and ousting features exclusive to Florence (among these are, probably, the change from the type lo mi dà ‘he gives it to me’, to me lo dà (Ch.3: 9.4.2), the triumph of the structure non facendolo ‘not doing it’ over non lo facendo (Ch.3: 9.4.1), the establishment of the type presero ‘they took’ over presono (Ch.3: 8.3.3), and other phenomena). Over the ensuing four centuries, the gulf between the literary language of Italy, and the speech of the Italians, tended to widen. Calculation of the proportion of the Italian people that could have been said to know Italian in the 1860s, at the time of the political unification of Italy, is fraught with difficulty.9 In so far as Italian was principally a written language, only the functionally literate – a minute proportion of the populace – were likely to be able to acquire a full command of the language. On the other hand, any native speaker of an Italo-Romance dialect, and particularly those who spoke Tuscan or another central Italian dialect, would have been able, given sufficient attention, to understand at least something of the Italian language, so that a degree of passive knowledge of Italian need not have been the exclusive preserve of the literate. Estimates of the numbers of those able to speak Italian at the time of Unification range between 2.5% of the population, according to De Mauro (1976), through 9.52% (or about two and a quarter million persons), according to Castellani (1982), to 12%, suggested by Serianni (1990: 18 n6). It should be added that a small proportion of Italians used speech varieties other than Italo-Romance. Indeed, their numbers increased after the First World War, due to border changes: De Mauro (1976: 10f.) estimates them at 2.1% of Italians (or about 800 000 souls) in 1921. In addition to the other Romance varieties (Ladin, Occitan, Franco-Provençal), there were also German dialects (spoken in Val Canale (Udine), parts of Trentino, and the Val de Gressoney (Aosta)), Slovenian (around Gorizia, Val di Resia and the upper Torre and Natisone valleys, and the area around Trieste), Greek (in clusters of villages in Salente to the south of Lecce and the Aspromonte area of Calabria), Albanian (scattered in villages in Abruzzo, Puglia, Basilicata, Calabria and Sicily), and Serbo-Croat (in a few villages in Molise). At Alghero in Sardinia a variety of Catalan was spoken. These speech communities all persist to this day.10 The perception of a linguistic gulf between the literary language and the speech of the Italian people, particularly where vocabulary is concerned, is keenly articulated in the nineteenth century by the Milanese Alessandro Manzoni, for whom the Questione della Lingua was no longer a debate about a literary language, but about the best means of extending knowledge of the language to the Italian people at large. But not only was Italian structurally remote from the everyday speech of most Italians, it was also functionally remote, in that it had remained elevated above the needs of everyday life, and was ill-equipped for use in everyday discourse. Part of Manzoni's solution was to propose that contemporary spoken Florentine (or a cultivated variety thereof), rather than archaic literary Florentine, should form the basis of the national language, and the final version of his novel I promessi sposi (1840) was intended, among other things, as an exemplar thereof. The culmination of a lifetime's meditation on this problem was Manzoni's report, commissioned by the Ministry of Education and published in 1868, on the unity of the Italian language and means of diffusing it, in which he proposed, inter alia, the teaching of Florentine in schools, and the publication of a modern Florentine dictionary. When the latter, even the first word of whose title – Novo vocabolario della lingua italiana secondo l'uso di Firenze – was distinctively Florentine (Ch.2: 4.2), began to appear in 1870, it provoked a memorable and insightful response from a scholar of linguistic history, Graziadio Ascoli. Ascoli (1873) cogently pointed out the impracticalities of a number of Manzoni's proposals, and above all that of overturning the established literary tradition in favour of a variety of Florentine many elements of whose structure were unfamiliar to the great majority of educated people. Ascoli also expressed his opposition to the imposition of inflexible linguistic models of any kind, whether archaic or modern. For him, the traditional literary language must be the basis of Italian, but its evolution as the national language of the Italians could only be the product of an increased and intensified cooperative intellectual activity on the part of the Italian people, a condition which he regarded as still lacking.11 In fact, political unification helped promote the expansion of Italian (the Italian of the literary tradition – not contemporary Florentine) along two major dimensions. The first was social: the Italian language was gradually acquired by the Italian populace at large. The mechanisms of this expansion, and the roles variously played by migration, military service, the educational system, the mass media, and other factors, have been explored, for example, by De Mauro (1976).12 In the late twentieth century, we find that the overwhelming majority of Italians understand and use Italian, although the way in which they use Italian may be divergent from the standard language based on the literary tradition, as we shall see. It is very difficult to obtain a true estimate of the proportion of the populace which remains substantially ignorant of the language, but we may safely say that it is minute, restricted to rural areas (especially of the far south and Sardinia), and to older generations. Perhaps the most telling indicator of the penetration of Italian among the Italians is whether they use the language at home, that is to say in the most intimate sphere of their lives. A recent survey (Doxa (1988)), in which a sample of Italians were asked to report on their own linguistic behaviour, suggests that about 34.4% of the population use only Italian, even in the home, while the remainder continue to use dialect at least with some members of their families, in addition to Italian. The proportion of those claiming to employ Italian at home (as well as in the wider world) at least some of the time was 60.4%. The second, accompanying, type of expansion concerns the domains of discourse in which Italian is used. As the language was acquired by the populace at large, so Italian was put to use in an ever widening range of domains, in casual conversation, in military life, in bureaucracy, in civil administration, and so forth. These expansions of Italian have resulted in structural diversity within Italian. The once remote and relatively homogeneous literary language has begun to show systematic variation according to region, social group, topic and context of discourse, and so forth. It is with such variation that shall be concerned in Chapter 5. 3 Chronology and historical sources 3.1 Chronology It is as difficult to divide Italian into chronological periods as it is to divide Italo-Romance dialects into geographical areas. From an ‘internal’ perspective we are inclined to think that such division is a counterproductive distraction. We shall not, then, be particularly concerned with periodization here, although we would not wish to deny the importance of this complex issue from a perspective which is primarily ‘external’. See particularly Weinapple (1983) and D'Achille (1991) for further discussion of problems of periodization. In this book we have preferred simply to state the (approximate) date for which the linguistic phenomena in question are attested. From an internal perspective, there is no historical moment of dramatic, structural upheaval which would allow us to distinguish ‘old’ and ‘modern’ Italian. Indeed, it is probably better for our purposes in this book to avoid the term ‘old Italian’ altogether, and to date the emergence of ‘Italian’, with quite deliberate vagueness, from the ‘fifteenth or sixteenth centuries’, the centuries during which a form of Tuscan dialect (based principally on Florentine) was becoming generally accepted as the literary language of Italy. In the history of ‘Italian’ thus defined, there is little need for further periodization. It would be a gross exaggeration to say that the structure of modern Italian had already assumed its present form by the fourteenth century. Indeed, we shall be at pains to underline the many changes – such as the rise of the progressive tense form of the type sto facendo ‘I am doing’, or the abandonment of interrogative structures of the type viene egli? ‘is he coming?’ (Ch.4: 3), etc. – which have occurred in Italian ever since the fifteenth century. But it is also true that Italian – primarily a literary idiom remote from spontaneous, everyday, language – has changed strikingly little since the time of Boccaccio and Petrarch, particularly in the domains of phonology and morphology. Before the fifteenth century, it might be more accurate to talk of ‘old Tuscan’ rather than of ‘old Italian’,13 since Tuscan was not universally accepted as the Italian lingua, even though the perception of the prestige and primacy of Tuscan was gathering ground from the time of Dante onwards. In the fifteenth century we find the terms fiorentino, toscano and italiano being used sometimes interchangeably. And, so far as one can determine, there was relatively little divergence between written Tuscan dialect and spoken Tuscan dialect. 3.2 Early texts The earliest concrete attestations of Italo-Romance are, of course, written texts. The first surviving continuous texts uncontroversially written in an Italo-Romance dialect, and not in Latin, are certain brief, formulaic, legal depositions dating from the 960s (the so-called Placiti cassinesi), written in a variety of Campanian.14 There is a hiatus of a century before the appearance, towards the end of the eleventh century, of the next Italo-Romance texts which have come down to us from Sardinia and central Italy. Romance texts (usually of a practical or religious nature) become increasingly frequent through the twelfth century, but our direct knowledge of OTuscan only commences with a text (a Pisan naval account register) written no earlier than the early twelfth century, while the first surviving text from Florence is a bankers’ book of 1211. It is only from the mid thirteenth century that Tuscan texts (including the first literary texts, for example, the Novellino) begin to appear with any frequency. 3.3 ‘Proto-Romance’ and ‘Classical Latin’ For our knowledge of the Romance of Italy as it emerged before the year 1000, we are dependent principally on reconstruction of ‘Proto-Romance’15 forms on the basis of comparison of the modern dialects, supported by our knowledge of their Latin antecedents, and of certain general principles of linguistic change. We cannot enter here into a critical discussion of the technique of comparative reconstruction. Suffice it to say that there are many cases where it is beyond reasonable doubt that a particular change in the structure of Latin underlies all or most of the modern Romance varieties of Italy, and that systematic differences between those varieties are wholly consistent with the postulation of a common historically underlying system; the evolution of the ‘western’ Romance vowel system, discussed in (Ch.2: 2.1) is a nice case in point. We follow in this book the convention of signalling forms whose existence is postulated, but unattested, by means of an asterisk; note that a double asterisk denotes some putative linguistic form whose existence is denied. The most convenient historical starting point, in terms of which to consider the subsequent evolution of Tuscan and other Italo-Romance dialects, is Classical Latin. But some caveats are in order. It is easy to fall into the trap of treating Classical Latin as if it represented a primordial state of linguistic neatness and cohesion which subsequently fragmented into the various Romance dialects. In reality, the Latin of the Roman Empire already displayed a vast range of variation according to region, social stratum, register of speech, and so forth. It is extremely unlikely that Classical Latin as represented by writers such as Vergil, Cicero or Caesar is the direct ancestor of Romance. While we are unconvinced by the hypothesis that Proto-Romance might have existed as a structurally distinct ‘sister’ of Classical Latin spoken in southern Italy already in the imperial period (cf. Hall (1950)), it seems very probable that Romance descends from the everyday, uncultivated, and already regionally variant, Latin of the Empire. That is to say that it has its origins in speech forms by no means wholly identical to those of Classical Latin. We shall take Classical Latin as the ‘starting point’ of our analysis, but it should always be borne in mind that we do this faute de mieux, because Classical Latin is probably the best approximation we have to the structure of the forebear of the Romance languages.16 4 A note on phonetic transcription In this book, where possible, we have given Latin and Italian words in their conventional orthography. When, as is often the case, this is inadequate for our purposes, we have transcribed into the International Phonetic Alphabet. However, certain phonological characteristics can be represented by means of minor adjustments to orthography. Latin orthography represents neither stress nor vowel length; we have signalled these by giving stressed syllables in bold type, and marking length with a colon (:) after the vowel. Stress in Italian words has been indicated, where necessary, by placing a grave accent over the stressed vowel (e.g., àmano ˈamano ‘they love’); where no stress is indicated, it should be assumed that the penultimate syllable is stressed (e.g., ama ˈanna ‘he loves’). As is conventional, stress in IPA transcriptions is indicated by the symbol1 immediately to the left of the stressed syllable (e.g., manˈdare). Particular attention should be drawn to a transcriptional practice followed in this book which deviates from standard practice, and reflects our methodological scepticism (explained in our comments on the ‘phoneme’ in section 5.4) about the value of the traditional distinction between ‘phonemic’ and ‘phonetic’ representations. We shall not use at all the slashes (/ /) normally associated with the representation of ‘phonemes’. Transcriptions of speech sounds will be indicated in bold face, and square brackets ([ ]), normally associated with ‘phonetic’ or ‘allophonic’ (and, therefore, ‘non-phonemic’) transcriptions. They will be used only where speech sounds are cited in isolation from the words in which they appear. In other words, ‘[e]’ means ‘this is the speech sound e (and not the letter ‘e’)’. 5 Some concepts in linguistic change Without attempting to provide even an outline of the theory and principles of language change (see, for example, McMahon (1994)), we sketch here, with some very brief illustrations, certain concepts which will be particularly useful in studying the history of Italian. Among these are • The inherent variability of language • The notion of ‘learnèd’ forms • Hypercorrection • The regularity and irregularity of sound change • Phonetically conditioned variation in speech sounds • Allomorphy • Grammaticalization (including morphologization) • Analogy • Emergence of analytic structures • Spoken vs. written language 5.1 Variation That language is spoken by human beings in human societies should be a statement of the obvious, but it is all too easy in charting the ‘internal’ linguistic history of a language, to talk as though the structural changes which have taken place were the result of ineluctable laws on a par with those of physics, operating on an abstract and homogeneous linguistic structure in which humans were merely the mindless conduit of change. Indeed, it is difficult not to talk in this way when discussing historically remote periods in the history of a language, of which our knowledge is, perforce, sketchy. But in reality, the way in which a speaker uses language may vary according to a wide range of factors (e.g., speed of speech, formality of the speech situation, and so forth); and social groups within a wider speech community may be distinguished by the systematically different ways in which their members use a common language. Speakers can be extremely sensitive even to minute patterns of variation, both within their own speech, and between social groups, and can choose to adopt, reject or modify aspects of the variation they hear about them. In short, no living language is homogeneous, and speakers are potentially active participants, not just passive automata, in the unfolding of linguistic change. Except perhaps for the inhabitants of the most remote and isolated localities, all Italians over the past two millennia must have been, at least, aware of the existence of varieties of language other than their own, ranging from the relatively minor perceptible speech differences between neighbouring villages, through the major differences in the language of Romance-speakers coming from more distant regions, and finally to Latin, and latterly Italian, as prestigious varieties of language superordinate to local dialect. Speakers who are aware of variation are also capable of imitating, often by adopting (‘borrowing’) into their own speech structural or lexical features of some other variety. We shall see that Florentine, for example, seems to have borrowed a variety of pronunciations and grammatical features from other Italo-Romance dialects – e.g., the pronunciation of ‘s’ as [z] between vowels in words like francese ‘French’ (Ch.2: 7.2) seems to have been taken from northern dialects, and the ordering of clitic pronouns me lo dà instead of OFlorentine lo mi dà, appears to be a matter of imitation of Sienese and of other dialects to the south of Florence (Ch.3: 9.4.2). The centuries of cultural prestige accorded to written Latin has meant that Latin has constituted a practically inexhaustible reservoir of ‘learnèd’ vocabulary for use in cultured discourse. Much of the learnèd vocabulary of Italian, i.e., words introduced into the language by an educated elite, is of Latin origin (Classical Greek is another major source), transmitted into Italian by intellectuals, clerics, lawyers, and others who possessed an active knowledge of Latin. Sometimes Latin words are adopted into Italian alongside the indigenous Italo-Romance words of which they are the historical source. This often has the effect of reintroducing into the language the historically underlying form of words which are already part of the native linguistic tradition. The word cosa ‘thing’ (from Latin CAUSA(M)) is the result of a continuous linguistic development going back to Roman times, displaying, for example, the characteristic monophthongization of the Latin diphthong [aw] (Ch.2: 4.4). But causa ‘cause’, is a learnèd borrowing into Italian from written Latin which, because it has not been continuously on the lips of Italians, has not undergone the sound changes to which its ‘indigenous’ counterpart has been subject. Such pairs of popular and learnèd words arising from a common source are termed etymological doublets. Imitation of prestigious linguistic models is not, of course, limited to the lexicon. 5.2 Hypercorrection Attempts to imitate varieties considered, for whatever reason, more prestigious or desirable, often result in hypercorrection: in their endeavours to reproduce the desired pattern of speech, speakers can, in effect, ‘overstretch’ themselves, producing novel structures which do not exist in the variety of speech which they are attempting to produce. We shall see that hypercorrection has played a major role in the linguistic history of Italian. An extreme, and potent, form of hypercorrection, is the exaggerated reinforcement of characteristic features of one's own speech as an assertion of local identity, while eschewing others which are in reality indigenous, but are perceived as ‘foreign’.17 Some changes in the history of Tuscan, including such historical puzzles as the unexpected appearance of the middle vowel [a] in cronaca ‘chronicle’ from Latin CHRONICA(M), or of [ηg] in parts of the verb tenere ‘to hold’ (e.g., tengo ‘I hold’), where we would expect to find [ɲɲ] (tegno), etc., are probably the results of hypercorrection (see Ch.2: 18.104.22.168 and Ch.3: 8.4.1). 5.3 Sound change and allomorphy Variation is inherent in the pronunciation of a language. Some variation is random and unsystematic, but if a variant articulation is systematically adopted by speakers, we say that a sound change has taken place. This is not the place to explore the intricate question of why sound changes occur, or how they spread through speech communities, but the adoption of variant pronunciation is often connected with a preference for relative ease of articulation. A potentially countervailing force is ease of perception: an easier articulation may obscure the identity of a speech sound, and so a potential change may be blocked, or changes may occur which serve to reinforce the identity of the intended sound. A nice example of both is the story of reduction and (hypercorrect) ‘restoration’ of unstressed vowels in Tuscan, discussed in (Ch.2: 22.214.171.124). Sound changes are not like physical laws, in the sense of blindly and automatically applying to all the sounds potentially subject to the change – although often, from the foreshortening perspective of history, it may look as if they have done so. Some sound changes do eventually operate on all occurrences of the sounds potentially subject to them but usually there is tension between a conservative norm and an innovatory pronunciation, and a speaker may apply the new pronunciation in some words but not in others, and in casual more than formal speech, and even in certain grammatical categories but not others, and so on. Some incipient changes may never become fully established because of the influence of competing linguistic varieties in which the relevant change does not occur, or occurs in a different way. An example is the story of palatalization by yod, discussed in Ch.2: 5.2, where ‘borrowing’ of the output of sound changes which have regularly taken place in other dialects has played a role. Probably the most helpful way to approach sound changes is to realize that they tend to affect every occurrence of the relevant sounds and in this respect are ‘regular’. But where those sound changes produce results which are systematically and perceptibly different from the pronunciation of other varieties of the language (for example, of more conservative speech patterns, or of the varieties spoken in other regions), there is a tendency for the influence of these coexistent and competing models to produce exceptions to otherwise general phonetic changes, and in this respect sound changes are often, perhaps usually, ‘irregular’. We shall pay particular attention in this book to the way in which sound changes have impinged on the grammatical structure of Italian. Some changes may result in a sound becoming identical to some other sound from which it was originally distinct: in this case we speak of merger. A very common scenario is that a sound changes only when a certain other sound precedes or follows it. Where this phonetic context is not present, the change does not take place. Contextually conditioned phonetic changes of this kind are usually assimilations – that is to say that the sound which changes adopts some phonetic characteristic of a neighbouring sound. A classic example in the history of Italo-Romance (Ch.2: 6), and of a great many other languages, is that velar consonants ([k] and [g]) tend to change into palatal consonants (articulated with raising of the tongue body towards the palate) in anticipation of the raised tongue position inherent in the articulation of a following front vowel [i] or [e]. Where no [i] or [e] followed, the change did not take place. This contextually conditioned split of one sound into different articulations has left its mark in the grammatical structure of Italian in the distinction between ‘diːtʃi ‘you say’, which displays the results of palatalization of an original [k] before a front vowel [i], and ‘diːko ‘I say’, where the velar consonant has remained intact, because a back vowel [o] follows. We can see that the sound change has impinged on the morphological system of Italian, in that the single meaning ‘say’, which in the Latin present tense was just diːk, is now represented by two forms, diːk in the first person singular, and diːtː in the second person singular. Where a particular meaning is expressed in this way by more than one form – a state of affairs which arises usually, but not necessarily, as a result of sound change – we speak of allomorphy. 5.4 A note on the ‘phoneme’ For methodological reasons, we have felt compelled in this book to deviate from certain usually held assumptions about the importance of the ‘phoneme’ in historical phonological description. This is not the place for a lengthy theoretical digression (for more detailed argumentation, see Maiden (1991)), but a few general remarks (which some readers may prefer to skip) are in order here. A point all too often neglected in historical linguistics is that we do not know at what point in the history of a sound change a resulting phonetic differentiation ceases to be merely a matter of more or less mechanical articulatory habit, and begins to be connected with the expression of the different grammatical categories with which the resulting differences are (at first accidentally) associated (e.g., first vs. second persons of the verb dire).18 There is no reason to identify this point – as many historical linguists of a structuralist bent do – with the moment at which a sound change becomes phonemic, that is when it is no longer exclusively and uniquely predictable from its phonetic context, and where the relevant sound difference can function on its own, independently of context, to distinguish the meaning of words which are in all other respects identical. Such minimally differentiated words are called ‘minimal pairs’, and constitute the principal diagnostic for phonemic opposition. For example, in modern Italian it is not the case that every [tʃ]19 can be derived from [k] preceding a front vowel (cf. cioccolato tʃ okkoˈlarto ‘chocolate’ where [tʃ] occurs before [o], and no [i] was ever present in the history of Italian), and not every [k] before a front vowel becomes [tʃ] (e.g., chilo ˈkrːdo ‘kilo’); moreover, [tʃ] vs. [k] can function alone, and independently of context, to distinguish the meaning of words: e.g., dici ʃdiːtʃi vs. dichi ˈdiːki, the former being the second person singular present indicative of dire, and the latter being a popular form of the present subjunctive of dire ‘to say’ (readers can find for themselves hundreds of other minimal pairs of this kind). Therefore, [tʃ] and [k] (or rather /tf/ and /k/, to present them in traditional phonemic representation, cf. section 4, above) are separate phonemes. When a once phonetically conditioned difference results in sounds which are phonemically distinct, and cannot be derived exclusively from their phonetic environment, we know that speakers must by now be associating the phonetic difference with differences of meaning. But speakers are perfectly capable of perceiving systematically occurring phonetic differences long before these cease to be exclusively predictable from their phonetic environment, and speakers are therefore perfectly capable of interpreting and using such phonetic differences – even while they are still motivated by articulatory factors – as additional expressions of associated differences in meaning. There is good evidence that this is precisely what does happen (cf. Maiden (1991)) and indeed we suspect that this is normally the case. Many linguistic histories make much of the distinction between phonetic (or ˈallophonic’) variation on the one hand and the emergence of phonemic opposition on the other. While the notion of distinctive opposition is of absolutely fundamental importance, it is far from evident that phonemic opposition defined in terms of minimal pairs deserves to be given much weight. We think that far less violence is likely to be done to the linguistic facts if the classification of sound differences into ‘phonemic’ and ‘allophonic’ is not overemphasized, and there will be little mention of the ‘phoneme’ in what follows. Overemphasis on the ‘phoneme’ defined in terms of minimal pairs can also be a distraction in another respect: a fertile imagination is needed to conceive of any circumstances under which the small handful of minimal pairs involving opposition between voiced and voiceless sibilants in Tuscan (such as ˈkjɛːse ‘he asked’ vs. ˈkjɛːze ‘churches’ (Ch.2: 7.2)), could function to distinguish meaning in normal discourse. The origins, and connotations for native speakers, of these distinct, voiced and voiceless, pronunciations are extremely interesting, as we shall see; but that minimal pairs may be found – with some effort on the part of the researcher – is much less so. 5.5 Morphologization and grammaticalization What is usually understood by morphologization is the integration of the effects of some sound change into the morphology of a language. ‘Morphology’ is concerned, in broad terms, with the relationship between the phonological shape of words and the meanings expressed by words: important factors in Romance morphology are the inflectional endings which express grammatical meanings, and the relationship between differences in the ‘stem’ (Ch.3: 1) of words, and the grammatical categories they express. We saw above how the results of palatalization of velar consonants can become part of the morphological person and number system of the verb, in addition to the inflectional endings. Another type of morphologization occurs when originally different words become the ‘same’ as far as lexical meaning is concerned, but are associated with different grammatical categories within the word: for example, we find and- in most persons of the verb andare ‘to go’, but va(d)- in the singular and third person plural forms of the present tense. These two forms were originally the stems of independent words, of different lexical meaning. Extreme allomorphy of this kind, where there is no phonological similarity between the alternating forms, is termed suppletion.20 Morphologization is a form of ‘grammaticalization’, but the term grammaticalization is commonly reserved for that development whereby an originally autonomous lexical word gradually loses its distinctive meaning, and becomes a mere indicator of grammatical categories. For example, the ending -mente, which in modern Italian is used to form the grammatical class of adverbs, originated as an independent word meaning ‘mind’, and the person and number endings of the future tense (e.g., finirò ‘I shall finish’, finirai ‘you will finish’), originate as present tense forms of the Latin verb HABERE, meaning ‘to have’, ‘to possess’. 5.6 Analogy Linguistic form often becomes reorganized in such a way that it constitutes a more perspicuous and diagrammatic expression of meaning. One mechanism by which this happens is termed ‘analogy’. The most prominent type of analogical change we encounter in the history of Italian occurs in morphology and takes the form either of analogical levelling, or of analogical creation of alternation. Allomorphy (see Ch.3: 0.2) tends to contradict the ‘normal’ expectation that each meaning will be signalled by just one linguistic form. That a single meaning should be expressed by two or more different forms involves, on the face of things, a wholly redundant complication of the grammar, and the relationship between form and meaning becomes ‘opaque’, while a one-to-one form-meaning relationship is maximally ‘transparent’. Where such allomorphic ‘complications’ are eliminated, by substitution of a single invariant form, where previously multiple forms indicated a single meaning, we speak of analogical levelling. For example, the OTuscan verb ‘to reap’ miètere displayed, as a result of sound change (Ch.2: 4.2), a diphthong wherever the (first) syllable was stressed (mièto ‘I reap’) but not where that syllable was unstressed (metèmo ‘we reap’). The discrepancy between the single meaning ‘reap’, and this duality of form, was resolved by generalizing the form containing the diphthong to all parts of the verb, whence modern Italian mièto, mietiàmo. Some grammatical differences may be expressed in certain parts of the grammar, but not in others. In OTuscan, the categories first and third persons singular in present tense, first conjugation, verbs were distinguished, respectively, by the endings -o and -a (parlo ‘I speak’ vs. parla ‘he speaks’); in the imperfect tense, however, the inflectional endings did not distinguish these persons (parlava ‘I spoke’ vs. parlava ‘he spoke’). Subsequently, on the analogy of the present tense, the -o vs. -a distinction was introduced into the imperfect (parlavo vs. parlava). In such cases we speak of analogical creation of alternation. Sometimes, this may even have the effect of disrupting one-to-one matchings between sound and meaning. For example, in OTuscan the verb fuggire had an invariant stem ːfudʤ-, meaning ‘flee’ (e.g., faggio ˈfudʤo ‘I flee’, fuggi ˈfudʤi ‘you flee’, etc.), but on the model of alternation between first person singular leggo ˈleggo ‘I read’, and leggi ˈlɛdʤi ‘you read’ (etc.), a new alternation has been created in fuggire, yielding fuggo ˈfuggo ‘I flee’ vs. fuggi ˈfudʤi ‘you flee’. Analogical changes can also operate in other parts of the grammar. For example, the extensive historical reorderings in the syntax of clitic pronouns in relation to the verb. In modern Italian we say lo vedo ‘I see it’ and vederlo ‘to see it’, but in OTuscan both vedolo and lo vedo and lo vedere or vederlo were, within certain constraints, possible. This change probably reflects an analogical generalization of the fact that in OTuscan the clitic tended to precede finite verbs, and tended to follow infinitives (Ch.3: 9.4.2). 5.7 The emergence of ‘analytic’ structures A major trend in the history of Italian (and of other Romance languages) is the emergence of so-called ‘analytic’ structures. Much of Latin morphology was ‘synthetic’, that is to say that the expression of lexical meaning (‘dog’, ‘do’, ‘tall’, etc.) was tightly ‘bound’, within a single word, with the expression of grammatical meaning (‘plural’, ‘genitive’, ‘first person’, ‘perfect tense’, etc.). Grammatical meaning was usually expressed by means of an inflectional ending inseparable from the verb, noun or adjective (e.g., CANIUM ‘of dogs’, FECI ‘I did’, etc.). In the emergence of analytic structures, the expressions of lexical and grammatical meaning get, so to speak, ‘unpacked’: in the history of Italian this has usually involved the assignment of grammatical meanings and lexical meanings to separate and originally autonomous words. Some of the functions of the Latin genitive inflections are taken over by a separate word, the preposition di (e.g., di cani ‘of dogs’), and some of the functions of the perfect tense are replaced by a structure comprising the verb avere ‘to have–, which bears information about person and number, and the past participle of the verb, which bears the lexical meaning of the verb (e.g., ho fatto ‘I have done’). What is particularly fascinating to trace in the history of Italian is the way in which the originally autonomous elements of the analytic structure (particularly in the verb) may become ‘grammaticalized’ (see above) in that they gradually lose the characteristics of semantically, syntactically and phonologically independent words, and become more and more like the ‘bound’ elements of the synthetic structures they originally replaced (e.g., Ch.3: 8.8). 5.8 Written and spoken language It is a fundamental principle of linguistic analysis that writing is secondary to speech. Suffice it to observe that the majority of languages that have ever been spoken have not been written down by their native speakers. And, until this century, the overwhelming majority of native speakers of Italo-Romance were illiterate, but no less fluent and accurate in their native speech for that. Virtually all the phonetic and other changes we chart in this book originated in the speech of the unlettered. Having made this clear, it is important to add that we should not underestimate the influence of writing in the history of Italian. For the last two thousand years, and probably longer, the Italians have lived in the shadow of the written word, not least because it was the medium of religious texts, and of legal and commercial language. Illiterate as the populace may have been, it is doubtful whether even the humblest peasant in the remotest village in the darkest moment of the Dark Ages did not know what reading and writing were, and did not recognize the prestige accorded to the written word. And in the Renaissance the proportion of literate males in the population sometimes reached surprisingly high levels in major urban centres, possibly higher than those reached in the nineteenth century.21 Where there is a discrepancy between speech and othography, there has long been a tendency among the literate to adapt pronunciation to spelling (‘spelling pronunciation’). And spelling has sometimes exercised a ‘distorting’ influence on the otherwise regular implementation of sound change. Since, for many centuries, the language of writing was par excellence Latin, and since Italian vocabulary has been continually replenished from written Latin, there have entered into Italian a great many Latin words pronounced, letter for letter, as they are spelled.22 This is one of the main sources of the rise of etymological doublets, such as causa ‘cause’ vs. cosa ‘thing’ (Ch.2: 13). The Roman alphabet, adapted for the writing of Latin, is ill-equipped to represent some of the sounds of Italian: for example, it offers no orthographic distinction between the open and closed mid vowels (Ch.2: 2.1), both spelled with ‘e’ and ‘o’, nor between [s] and [z] (both spelled ‘s’) and [ts] and [dz] (both spelled ‘z’).23 Such ambiguities in the relation between sounds and letters have played a role in the emergence of ‘popular’ Italian pronunciation in this century (Ch.5: 3.1.3), with speakers failing to make those distinctions which are not unambiguously indicated in standard Italian orthography, but also sometimes pronouncing letters which in modern Italian have no phonetic value (such as the ‘i’ in giallo ‘yellow’ – where ‘i’ is a ‘diacritic’ sign, indicating that the preceding ‘g’ is pronounced [ʤ], not [g], or in cielo ‘sky’, where the ‘i’ is not pronounced in modern standard Italian). There are other respects in which writing and speech are different. Spoken language is usually more spontaneous and less carefully meditated than written language. Written language is permanent, and is usually that medium in which norms of ‘correct’ linguistic usage are first laid down. Spoken language tends to favour innovation in linguistic structure, whereas the written word is more conservative. The permanency of writing allows the reader to look back over what has been written, and this fact facilitates the use of syntactically far more complex structures than tend to be used in spoken language. Complex syntactic structures in spoken language tend to place a heavy burden both on the speaker's and on the hearer's memory, and simpler structures tend, accordingly, to be preferred. We shall see many examples in what follows of structural differences between written and spoken Italian, ranging from the use of personal pronouns (Ch.3: 9.1.2), to the basic ordering of elements in the sentence (Ch.4: 1). Notes 1. See Varvaro (1984: 25–7; 33f.; 38–40) for an analysis of the history of this distinction. 2. On this point see also Varvaro (1984: 45–7). But the present work is closer in kind to what he describes as a ‘sophisticated form of structural diachrony’ than to his conception of a ‘history of the language’. 3. To mention but a few, there are Migliorini and Griffith (1984); Bruni (1987); and the new multivolume series Storia della lingua italiana, emerging under the general editorship of F. Bruni, which combines rich textual documentation with detailed linguistic analysis. A succinct and up-to-date account of the rise of Italian as a ‘standard’ language since the Renaissance appears in Muljačić (1988). 4. This definition clearly includes Sardinian and Friulian, but excludes Franco-Provençal, for which the ‘guiding’ language is French. Unfortunately, it also tends to exclude Corsican, which is patently a variety of Tuscan (introduced to the island around 1000), yet most of whose speakers would now recognize French, not Italian, as their ‘guiding’ language (see also Dalbera-Stefanaggi (1991: 27–30)). 5. Substratist accounts are on much firmer ground where the lexicon is concerned. Words such as verna ‘elm’ and brüc ‘heather’ in northern dialects are known to be of Celtic origin. The attrufu ‘October’ of Basilicata can be shown to be of Oscan origin. See Rohlfs (1972b: 17–19). 6. For an account of early flowerings of Italo-Romance dialects as literary languages, see, for example, Migliorini and Griffith (1984, Chapter 4). 7. For a historical account see particularly Vitale (1984), and the succinct account in Hall (1942a). Menocal (1991) gives a sober assessment of the importance of the Questione for the history of Italian. 8. It is likely that the structural conservatism of Florentine, making it apparently more similar to Latin than are other Italo-Romance varieties, may also have served to promote its acceptance as a worthy successor to Latin. 9. See, however, Pellegrini (1975b: 64f.). 10. For further details see, for example, Bruni (1987: 287f.). Note that Val d'Aosta has two official languages, Italian and French. The number of German speakers increased greatly through the incorporation of the Austrian Südtirol (Alto Adige) after the First World War. 11. For the nineteenth century Questione della Lingua, and especially the debate between Manzoni and Ascoli, see Vitale (1984), also Serianni (1990:41–67). 12. See also Serianni (1990: 23–6). 13. If the term ‘old Italian’ can be used at all, it is in the essentially geographical sense of all the various dialects used in the Middle Ages in the territory of what is now Italy, and in which Italian is now used and accepted as standard language (see also Varvaro (1984: 48 n155)). 14. For analysis of the earliest Italo-Romance texts see, for example, Castellani (1973). 15. Since there was never any break in continuity of transmission between the spoken Latin of Italy in the Imperial period, and modern Italo-Romance, we see little advantage in distinguishing ‘Vulgar Latin’ from ‘Proto-Romance’, or in trying to identify the point when Latin ‘ended’ and Romance ‘began’. 16. A very useful introductory sketch of Latin for Romance linguists is Vincent (1988b). 17. For a discussion of the type of concepts mentioned here, see, for example, Hudson (1980: 195–201) and Chambers and Trudgill (1980: 87f.). 18. The same arguments given here can apply to cases of ‘lexicalization’, where the result of some sound change becomes analysed as a primary element of the phonological shape of a word: the initial palatal [tʃ] of cena ːtʃeːna ‘dinner’ is the result of historical palatalization rules affecting an original velar consonant *[k] before a front vowel. But nothing rules out the possibility that, at some stage in its history, the palatal consonant could be both the Șallophonic’ result of a ‘live’ phonetic process of palatalization, and part of the lexical representation of a word. In other words, we do not accept that the phonological representation of words must be ‘phonemic’. 19. See section 4 above for the value of square brackets. 20. Similar to morphologization is lexicalization, discussed in note 18. 21. For an account of the extent of literacy in early modern Italy, see Burke (1987). 22. It may not always have been so. Wright (1982) develops the stimulating and still controversial thesis that, until the time of Charlemagne, in France, and the twelfth century in Spain, written Latin and spoken Romance were not perceived as separate languages, and Latin was not pronounced ‘one sound for one letter’. Rather, written Latin was simply the writing system of spoken Romance. In principle, it might also have been the case in the Italy of the first millennium that, for example, one wrote UIRIDIS, but said ˈverde ‘green’, and perhaps even wrote IGNIS, but said ˈfwɔːko ‘fire’. What is certain is that there were some respects in which written Latin was adapted to native pronunciations in Italy: Latin CT was generally pronounced [tt], TI + vowel was pronounced [tsj] + vowel, and GN was pronounced [ɲ] (e.g., SECTA > ˈsɛtta setta ‘sect’; UITIUM > ːvittsjo vizio ‘vice’; IGNORO > iɲˈɲɔːo ignoro ‘I am unaware’). 23. For an account of attempts, in the sixteenth century, to distinguish these sounds orthographically, see Migliorini and Griffith (1984: 228–30). The orthographical treatises of Trissino and others are gathered in Richardson (1984). For subsequent attempts at spelling reform, see Migliorini and Griffith (1984: 323; 374; 423). Chapter 2 History of the sound system 0 Introduction In the development of the the Italian sound system from Latin, the losses have been few, the innovations many. Some of the latter, notably the enrichment of vowel quality distinctions, and the emergence of palatal consonants, have had a major impact on the morphological structure of Italian. For this reason, the following account of Italian historical phonology has a strongly morphological bias, in that it explores in particular those changes which have helped to confer on the language its characteristic morphological shape. Sections 1 to 3 of this chapter summarize the principal phonological developments from Latin1 to Italian. Thereafter, a number of sound changes which have had a major impact on morphological structure will receive a more detailed investigation. 1 The prosodic system Prosodic (or suprasegmental) phonology is concerned with domains or ‘stretches of sounds’, greater than the single consonant or vowel segment. Among these are the syllable, stress (the relative loudness of syllables within some domain – usually the word), and length (concerning the relative duration of a vowel or consonant). If we conceive the chain of speech sounds as comprising a series of segmental ‘slots’, then long vowels or consonants might be viewed as those whose articulatory features are spread over more than one such ‘slot’.2 1.1 The syllable Of the syllable we need only say, for the moment, that in CL, as in modern Italian, it always comprises a nucleus constituted by a vowel, which may be preceded by a consonantal onset, and may be followed by a consonantal coda. Clusters of up to three consonants are admitted in the onset. The last consonant of an onset cluster must be a liquid ([r] or [l]), the first a sibilant ([z] or [s]), and the middle one an obstruent.3 In the coda up to two consonants are admitted, but the first must be either a liquid or a nasal. Some examples are: SPLEN-DET ‘it shines’; STRIC-TUS ‘squeezed’; MULC-TUS ‘milked’; PROMPT-SI ‘I produced’; FARC-TUF ‘stuffed’. The preferred syllable structure, in CL as in Italian, and as in a great many other of the world's languages, is open, which is to say that there should be no consonant in the coda. This preference is apparent in the fact that, in sequences of a single consonant followed by a vowel (or by a liquid, [r] or [l], followed by a vowel), the consonant always forms a syllable with the following vowel, never with a preceding vowel (e.g., CL A-MA-BA-TIS ‘you loved’; PA-TREM ‘father’ and not **AM-AB-AT-IS; **PAT-REM/**PATR-EM. We show in sections 8 and 9 below that such preferred syllable structures have played a considerable role in various consonantal changes in Italian. 1.2 Stress In CL, stress was proparoxytonic (i.e., it fell on the the last syllable but two of a word of three or more syllables), unless the penultimate syllable was heavy in which case stress was paroxytonic (i.e., it fell upon the penultimate syllable of a word). The definition of a heavy syllable is that it contains either a long vowel or a vowel, whether long or short, followed by a consonant. The operation of stress placement can be conveniently illustrated from the following morphological forms (1): (1) COL -LI-GO: ‘I gather’ COL-LEC -TUM ‘gathered’ COL-LE: -GI: ‘I gathered’ A -MAT ‘he loves’ A-MA: -BAT ‘he loved’ A-MA-BA: -MUS ‘we loved’ A -MANT ‘they love’ A-MAN -TUR ‘they are loved’ DO -MI-NA ‘mistress’ DO-MI-NA: -RUM ‘of mistresses’ POR -TA ‘door’ POR-TA: -RUM ‘of doors’ As a rule, the stressed syllables of Latin words4 are continued as stressed syllables of Italian words,5 although their relative position in the word may have altered as a result of other changes. Syncope (the deletion of non-final unstressed vowels, discussed in 126.96.36.199), has made COLLIGO and DOMINA into paroxytones (colgo; donna). A stress pattern almost unknown6 to Latin is oxytonic (last syllable) stress, which has arisen in the history of Italian usually as a result of deletion of a final vowel or a final syllable: AMAUIT > Proto-Romance *aˈmaw(t) > amò ‘he loved’; CIUITATE(M) > OTuscan cittade > città ‘city’; UIRTUTE(M) > OTuscan virtude > virtù ‘virtue’. 1.2.1 Stress in Italian Latin stress placement was entirely predictable on the basis of phonological information. The position of Italian stress is, in contrast, often impossible to predict without reference to non-phonological information. Practically all that remains of the Latin system is the constraint debarring placement of stress to the ‘left’ of the antepenultimate syllable (and even this is violated by certain third person plural verb forms, such as telèfonano ‘they telephone’, which will be analysed further in Ch.3: 8.3.3). Otherwise, stress may fall on any syllable from the antepenultimate rightwards, regardless of syllable structure. This new freedom of stress (shared, historically, by all Romance varieties) has allowed the emergence of meaningful distinctions between stress patterns, such as presentò ‘he presented’ vs. presènto ‘I present’, or càpito ‘I turn up’; capito ‘understood’ and capitò ‘he turned up’. However, stress is generally not admitted to the left of the penultimate syllable if that syllable is closed (i.e., if it ends in a consonant). Thus we have presentò and presànto, but not **prèsento. This is a remnant of the CL ‘heavy penultimate’ rule and, while it is indisputable that the overwhelming majority of Italian words happen to conform to this stress, and that modern Italians tend to assume that this rule operates when they pronounce unfamiliar words, it is questionable whether, in the modern language, this rule constitutes a particularly powerful constraint on stress-placement. After all, violations of this principle are found in what are, historically, loanwords (especially those originally taken from Greek), and in certain placenames (màndorla ‘almond’; pòlizza ‘insurance policy’; còrizza ‘coryza’; àrista ‘chine of pork’; Tàranto; Lèpanto; Àgordo) but these show no sign of being corrected in favour of the ‘normal’ pattern. Italian displays a relationship between stress and syllable structure which is an inversion of that obtaining in CL. Whereas CL stress was predictable partly on the basis of syllable structure, syllable structure in Italian (as, historically, in other Romance varieties) becomes partly predictable from stress, and from certain other aspects of word structure. The modern Italian rule is that all stressed syllables must be heavy. An important qualification is that open stressed syllables in oxytones are light before a pause or a vowel (par-ˈlɔ parlò ‘he spoke’; par-ˈlɔ-ˈaɲ-ke parlò anche ‘he also spoke’). Open syllables which were not originally heavy have become heavy by means of ‘segment lengthening’: ˈaː-ma-no amano ‘they love’; ˈfaː-tʃi-le facile ‘easy’; de-ˈtʃiː-ze-ro decisero ‘they decided’, etc. The formulation ‘segment lengthening’, rather than the more traditional ‘vowel lengthening’, is deliberate, since we shall be suggesting in 8.2 that both lengthening of vowels and lengthening of consonants were historically available to make up the ‘weight’ of a stressed syllable, and that the type of historical consonant lengthening detectable in proparoxytones such as fèmmina ‘woman’ < FEMINA(M), or after the stressed syllable of oxytones when a consonant follows (e.g., par-ˈlɔ parlò, but par-ˈlɔb-ˈbɛː-ne parlò bene ‘he spoke well’) represents an alternative strategy for the formation of heavy stressed syllables. We shall see below that vowel length, once a partial determinant of stress placement, is now exclusively determined by stress and syllable structure, with long vowels appearing solely in stressed open syllables. 1.3 Length Distinctions of length (or ‘quantity’) played a central role in the CL vowel and consonant system, and minimal pairs, distinguished only by the opposition between long and short vowels or consonants, were common (e.g., POPULUS ‘people’ vs. PO:PULUS ‘poplar’; FRICTUS ‘rubbed’ vs. FRI:CTUS ‘fried’; GRADUS ‘step’ vs. GRADUS ‘steps’; FALLAM ‘I shall deceive’ vs. FALAM ‘scaffolding’; AGER ‘field’ vs. AGGER ‘embankment’, etc.). Unlike modern Italian, vowels could be distinctively long or short, irrespective of the structure of the syllable in which they occurred (cf. Allen (1978: 64–77)). Not only are CL long consonants generally preserved intact in Italian (MITTO ‘I send’ > metto ‘I put’; PASSU(M) ‘step’ > passo, etc.), but some originally short consonants have become lengthened (e.g., HABEAT ‘let him have’ > abbia), in circumstances which will be explored in 8.2. Preservation of distinctive consonant length is a characteristic which sets Italian (and other central and southern Italian dialects, together with Sardinian), apart from all other modern Romance varieties. Vowel length evolves quite differently. CL differences in vowel length have been lost, but not necessarily without trace, since they have usually been transformed into differences of vowel aperture – see 2.1). At some stage after the loss of CL vowel length, there appeared a new, phonologically determined, differentiation of vowel length, which is continued into modern Italian, and underlies most Italo-Romance and Gallo-Romance varieties. The new distributional principle for vowel length is that all vowels are short, unless they occur in a stressed and open syllable, in which case they are long. These length variations are especially apparent in the verb stem, which may be stressed or unstressed (Ch.3: 8.1), and where, in some cases, an open-syllable stem may alternate with a closed-syllable stem. Thus the third person present indicative, passato remoto, and imperfect indicative forms of the modern Italian verbs dire ‘to say’; valere ‘to be worth’; tradurre ‘to translate’ are (2): (2) Stressed Open Syllable Stressed Closed Syllable Unstressed Syllable ˈdiː-tʃe dis-se di-ˈtʃeː-va ˈvaː-le ˈval-se va-ˈle:-va tra-ˈduː-tʃe tra-ˈdus-se tra-du-ˈtʃeː:-va This distinction between the long vowels of stressed open syllables and the short vowels of all other syllables plays a particularly important role in the history of Gallo-Romance and Italo-Romance vocalism, since the long vowel of open syllables is often subject to further differentiation (particularly diphthongization) with respect to its short counterparts. A putative case in point, whose nature and origin will be reviewed at length in 4.2, are the diphthongs [jɛ] and [wɔ], which have arisen in stressed open syllables in Italian, as in many Gallo-Romance dialects (e.g., Italian muore ‘dies’ vs. morto ‘dead’, moriva ‘died’; siede ‘sits’ vs. seduto ‘sat’; uomo ‘man’ vs. omino ‘little man’, etc.). 2 Vowels 2.1 Loss of vowel length, and expansion of aperture distinctions The CL vowel system comprises five vowels, each of which can be distinctively long or short, so that there is a ‘5×2’ vocalic system (3): Latin vowel length has been lost in all Romance varieties. At one extreme, Sardinian (supposedly with certain dialects of southern Basilicata) has lost vowel length distinctions without trace, with long and short vowels simply merging: e.g., MU:RU(M) ‘wall’ > ˈmuːru, CRUCE(M) ‘cross’ > ˈruːke, SO:LA(M) ‘Alone’ > ˈsoːla, ROTA(M) > ˈɔːta, CARU(M) FEL ‘gall’ > ˈfɛːle, UE:RA(M) ‘true’ > ˈvɛːra, NIUE(M) ‘snow’ > ˈniːve, FI:LU(M) ‘thread’ > ˈfiːlu, etc. The vowels of Romanian (and, apparently, of a tiny pocket of southern Basilicata, near Potenza) have followed the ‘Sardinian’ development so far as [a], and back vowels, are concerned. The front vowels of Romanian continue to distinguish original long and short realizations, as in Italian.7 Italian, in common with the great majority of Romance dialects, has lost only the distinction between Latin long and short [a]. The other vowel length differences have been transformed into differences of vowel quality, with an attendant expansion of the range of vowel qualities. Latin short vowels were pronounced with a more open articulation than their long counterparts; (cf. Allen (1978: 48f.)). In the vowel system historically underlying Italian, the short vowels (other than the already maximally open [a]) enhanced these aperture distinctions, so that, while long [iː] and [uː] retained their quality, short [i] and [u] opened to [e] and [o], thereby merging with the reflexes of Latin long [eː] and [oː]. Short [e] and [o], however, assumed a low mid articulation ([ɛ] and [ɔ]), in stressed syllables. The result was a system with seven vowels (4): These changes can be seen in: MU:RU(M) ‘wall’ > ˈmuːro, LU:CTU(M) > ‘mourning’ > ˈlutto, UO:LUIT ‘he wanted’ > volle, SURDU(M) ‘deaf > ˈsordo, *ROTA(M) ‘wheel’ > ˈrwɔta, PORTO ‘I bear’ > ˈporto, CA:RU(M) ‘dear’ > ˈkaːro, CAPUT ‘head’ > ˈkaːpo, UENIT ‘he comes’ > ˈvjɛːne, PERDIT ‘he loses’ > ˈpɛrde, UE:NIT ‘he came’ > ˈvenne, DICTU(M) ‘said’ > ˈdetto, DI:CIT ‘he says’ > ˈdiːtʃe, UI:XIT ‘he lived’ > ˈvisse. The same seven-vowel system probably underlies the modern vowel system of Sicilian (with Calabrian and Salentino), where [e] and [o] have been further raised, to merge with [i] and [u] (e.g., Sicilian ˈmuːru, ˈluttu, ˈsurdu, ˈdiːtʃi, ˈvissi, ψditto). 2.2 Loss of Latin diphthongs Of the three Latin diphthongs, Æ [ai], CE [oi], AU [aw], only the last survived into Italo-Romance (its later monophthongization as [ɔ] in Italian will be discussed in 4.4). All three were already subject to occasional monophthongization in the classical period, especially in rustic speech (cf. Allen (1978: 60–2)). In Proto-Romance the diphthongs [ai] and [oi] had merged, respectively, with reflexes of Latin short [e] and long [eː]. Thus, CÆLU(M) > ˈtʃɛːlo ‘sky’, and PŒNA(M) > ˈpeːna ‘grief’. 2.3 Stress and vowel quality The Latin ‘5×2’ vowel system has been resolved as a seven vowel system only in stressed syllables. In unstressed syllables, [ɛ] and [ɔ] do not appear. Rather, short O and E merge with their long counterparts, to give [o] and [e]: e.g., porTAT and PORTATIS > ˈpɔrta ‘he carries’, but porˈtarte ‘you carry’; PERDIT and PERDEBAT > ˈpɛrde ‘he loses’, but perˈdeːva ‘he lost’. The result is a system of five unstressed vowels: [u] [o] [a] [e] [i]. This differential development of stressed and unstressed vowels is amplified in the subsequent history of Tuscan: while the vocalism of stressed syllables is enriched by the emergence, in stressed open syllables, of the diphthongs [jɛ] and [wɔ] from [ɛ] and [ɔ] (see 4.2), the range of unstressed vowels tends to be further reduced to four or even three, in a variety of ways, to be examined in 4.6.1. 3 Consonants 3.1 Loss of [h] The list of phonological ‘losses’ in the evolution of the consonant system is brief indeed. The one complete loss is that of Latin [h]. This sound, already subject to deletion in the classical period (cf. Allen (1978: 43f.)), has disappeared without any trace in all Romance languages. Thus, in Italian, HABEBAT > aveva ‘he had’; PER HOC > però ‘however’, etc. The consonant [h] may have been reintroduced in later loans from Germanic, but even so, it subsequently disappeared from all Italo-Romance varieties: e.g., *ˈhaɲka > anca ‘hip’; *ˈhaspa > aspa ‘reel’. It does not appear in learnèd borrowings from Latin (or Greek): e.g., HORRIBILEM ‘horrible’ and HABITARE ‘to dwell’ are borrowed as orˈriːbile and abiˈtare. 3.2 Emergence of the voiced fricative [v] Latin possessed the voiceless fricative [f], but lacked its voiced counterpart [v] (fricatives are consonants uttered by expelling air between speech organs, in this case the lips and the teeth, which are brought close together, thereby producing turbulence). There are two major sources of [v] in Romance. One of them is Latin [w] in syllable-initial position (e.g., UIUO > vivo ‘I live’ – a development to be discussed in 8.1); the other is Latin word-internal intervocalic [b], which gave [v] in common Proto-Romance, probably via the fricative [β]: FABA(M) > fava ‘bean’; HABEBAT > aveva ‘he had’; DEBET > deve ‘he must’. In the perfect tense of the verb, however, intervocalic U was deleted; AMAUI > amai ‘I loved’; AMAUISTI > amasti ‘you loved’. Latin word-initial [b] remains intact in Italian; barba ‘beard’; battere ‘to beat’; bene ‘well’; bere ‘to drink’, etc., are examples. The Italian forms bevo ‘I drink’ < BIBO; vivo ‘I live’ < UIUO and serva ‘maid’< SERUA(M); erba ‘grass’ < HERBA(M), illustrate how the outcomes of Latin [b] and [w] coincide in word-internal intervocalic position, but remain distinct elsewhere. There are occasional occurrences of [rb], rather than [rv]: NERUU(M) > nerbo ‘sinew’; SERUAT > serba ‘he keeps’. 3.3 Postconsonantal [l] > [j] In virtually all Italo-Romance varieties, the liquid consonant [l] has become [j] when preceded by a tautosyllabic consonant (i.e., by a consonant that occurs in the same syllable – cf. 8.2) (5). (5) Proto-Romance Italian *ˈplenu ˈpjeːno ‘full’ *ˈtɛmplu ˈtɛmpjo ‘temple’ *ˈdoplu ˈdoppjo ‘double’ *ˈblaŋku ˈbjaŋ-ŋko ‘white’ *ˈflore ˈfjoːre ‘flower’ *korŋflare gorŋˈfjaːre ‘swell’ *ˈklaru ˈkjaːro ‘clear’ *ˈkave ˈkjaːre ‘key’ *ˈglanda ˈgjanda ‘acorn’ The cluster [sl] is resolved as [skl], which then yields [skj]: *ˈslavu > *ˈsklavu > schiavo ‘slave’. The incidence of original consonant + [l] sequences is augmented by syncope of unstressed vowels (see 188.8.131.52), notably in the originally diminutive suffix -ULUS. Note that *[tl] > [kl]: (6) CL Proto-Romance Italian FIBULA(M) *ˈfibla ˈfibbja ‘brooch’ OCULU(M) *ˈɔklu ˈɔkkjo ‘eye’ UNGULA(M) *ˈuŋgla ˈuŋgja ‘nail’ UETULU(M) *ˈvɛklu ˈvɛkkjo ‘old’ The change of postconsonantal [l] to yod is relatively minor, since it does not give rise to morphological alternations. It attracts attention because it comes close to constituting the sole common and distinctive characteristic of Italo-Romance dialects. But the phenomenon is also represented, in various phonetic forms, outside Italy (although only in Italo-Romance is the palatal realization of [l] attested after all preceding consonants in the same syllable); and within the Italo-Romance domain, there are dialects of north-eastern Lombardy, and of Abruzzo where postconsonantal [l] survives. In the relevant parts of Abruzzo, while the change has taken place in the original clusters [kl] and [gl], [l] has been maintained after other consonants: e.g., Bellante cɛf ‘key’ < *ˈklave; ˈviccə ‘old’ < *ˈveklu; blɛŋk ‘white’ < *ˈblanku. The mechanism of this development is explored in detail, from a comparative Romance perspective, by Repetti and Tuttle (1987), who argue that the change of postconsonantal  to [j] (via a postulated stage *[λ) was originally limited to consonant clusters comprising velar + lateral (i.e., just [gl] and [kl]). This initial phase of the development is conserved in some modern varieties of Abruzzese and is apparent in OPaduan and medieval Romagnol and Marchigiano. Elsewhere, it extended to other consonantal environments. The intervocalic cluster [gl] has sometimes yielded [λλ], rather than [ggj] in Tuscan: STRIG(I)LE(M) > ˈ;striλλa ‘currycomb’; UIG(I)LARE > veλψλaːre ‘to stay awake’; TRIGLA(M) > ˈtriλλa ‘red mullet’; TEG(U)LA(M) > ˈteλλa ‘baking tray’ (also ˈteggja). Castellani (1954) assumes that the original Tuscan development was [ggj], and that the emergence of [λλ] is a hypercorrection motivated by the merger, in less prestigious Tuscan varieties, of both [λλ] and [ggj], yielding the voiced palatal affricate [ɟɟ] (e.g., ˈfiɟɟa < ˈfiλλa ‘daughter’). On this account, speakers purportedly tended to ‘correct’ every incidence of [ɟɟ] in favour of [λλ], regardless of its origin. A different view is that of Repetti and Tuttle (1987: 81), who opine that [λλ] is the result of phonetic changes affecting [g] before [λ]. 3.4 Emergence of the palatal and affricate consonants Palatal consonants are those articulated by making contact between the body of the tongue and the hard palate; affricate consonants are those produced by completely blocking the flow of air through the mouth, then gradually releasing the air, so that friction, or ‘turbulence’ is produced. Although rather different in nature (‘palatal’ refers to a point of articulation, and ‘affricate’ to a manner of articulation), it is appropriate to discuss them together here because of the common elements in their origin in Italo-Romance. Latin had neither type of consonant, although it is probable that, before a vowel, what was written ‘I’ was pron